From: Paul Sandoz <Paul.Sandoz_at_Sun.COM>
Date: Thu, 03 Dec 2009 13:19:13 +0100

On Dec 3, 2009, at 1:00 PM, Martin Matula wrote:

> Hi,
>
> On Dec 3, 2009, at 12:26 PM, Paul Sandoz wrote:
>
>> Hi Micha,
>>
>> The current method of contextual encoding is by design in JAX-RS.
>> Rather embarrassingly the case you describe never occurred to me or
>> others on the EG.
>>
>> Before we modified UriBuilder to support contextual encoding we had
>> encode(true/false) and isEncoded methods on UriBuilder. The
>> contextual encoding approach simplified the API.
>
> I think the API is good. The problem is the encoding routine does
> not encode "%30" properly - i.e. the following is false:
>
> decode(encode(string)).equals(string)
>
> Isn't that a bug in the URIComponent._encode() method?

Not if contextual encoding is enabled, which is the type of encoding
that is used by UriBuilder.

Try this:

         String s =
UriComponent.decode(UriComponent.contextualEncode("some is %30 of X",
Type.QUERY),
                 Type.QUERY);
         System.out.println(s);

Compared to this:

         String s = UriComponent.decode(UriComponent.encode("some is
%30 of X", Type.QUERY),
                 Type.QUERY);
         System.out.println(s);

Paul.